检索结果-内蒙古大学图书馆

A stock market trading framework based on deep learning architectures

MULTIMEDIA TOOLS AND APPLICATIONS 2022年第10期81卷 14153-14171页

作者： Shah, Atharva Gor, Maharshi Sagar, Meet Shah, Manan Nirma Univ Dept Mech Engn Ahmadabad Gujarat India Software Engineer Quinbay Technol Bangalore Karnataka India Data Sci Associate ZS Pune Maharashtra India Pandit Deendayal Petr Univ Sch Technol Dept Chem Engn Gandhinagar India

Market prediction has been a key interest for professionals around the world. Numerous modern technologies have been applied in addition to statistical models over the years. Among the modern technologies, machine learning and in general artificial intelligence have been at the core of numerous market prediction models. deep learning techniques in particular have been successful in modeling the market movements. It is seen that automatic feature extraction models and time series forecasting techniques have been investigated separately however a stacked framework with a variety of inputs is not explored in detail. In the present article, we suggest a framework based on a convolutional neural network (CNN) paired with long-short term memory (LSTM) to predict the closing price of the Nifty 50 stock market index. A CNN-LSTM framework extracts features from a rich feature set and applies time series modeling with a look-up period of 20 trading days to predict the movement of the next day. Feature sets include raw price data of target index as well as foreign indices, technical indicators, currency exchange rates, commodities price data which are all chosen by similarities and well-known trade setups across the industry. The model is able to capture the information based on these features to predict the target variable i.e. closing price with a mean absolute percentage error of 2.54% across 10 years of data. The suggested framework shows a huge improvement on return than the traditional buy and hold method.

关键词： Convolutional neural network (CNN) Long short term memory deep learning architecture

来源：评论

学校读者我要写书评

暂无评论

Performance evaluation of deep learning architectures for load and temperature forecasting under dataset size constraints and seasonality

引用

ENERGY AND BUILDINGS 2023年第1期288卷

作者： Choi, Wonjun Lee, Sangwon Chonnam Natl Univ Sch Architecture Gwangju 61186 South Korea Dartwork Seoul 07503 South Korea

Buildings and their energy systems have unique characteristics and temporally changing dynamics. Additionally, building data have strong seasonality. Therefore, a deep learning model with good adaptability (i.e., high performance with little data) is necessary for model predictive control of buildings. Considering the specificity of the building-energy field, we benchmarked six deep learning architectures: the multilayer perceptron, simple recurrent neural network (RNN), long short-term memory, gated recurrent unit (GRU), dilated convolutional neural network, and transformer. Moreover, the data similarity analysis method was developed to analyze the effect of data seasonality on forecasting performance. A publicly accessible data generator and the open-source Python library deepTimeSeries were used to design a reproducible benchmark. The zone temperatures and thermal loads over a future 24-h period were the forecasting targets. The benchmark results, with varying training dataset sizes from 0.3 to 0.9 y, demonstrated that the transformer architecture performs best, especially on small training datasets. The GRU and RNN were the second- and third-best performers, whereas the rank of other architectures varied depending on the training dataset size and forecasting targets. The similarity analysis revealed that simply increasing the dataset size does not necessarily improve performance, indicating the importance of training with highly similar data for the forecasting period.(c) 2023 Elsevier B.V. All rights reserved.

关键词： deep learning architecture Time -series forecasting Transformer recurrent neural network (RNN) dilated convolutional neural network (DCNN) model predictive control (MPC)

来源：评论

学校读者我要写书评

暂无评论

Enhancing low radio-dosage thyroid imaging using deep learning and Monte Carlo-based absorbed dose measurement in nuclear medicine

引用

NUCLEAR ENGINEERING AND TECHNOLOGY 2025年第5期57卷

作者： Kim, Hochul Park, Chanrok Eulji Univ Dept Radiol Sci 553 Sanseongdaero Seongnam Gyeonggido South Korea

The acquisition of improved quality via the administration of low radioactivity is an important challenge in the field of nuclear medicine. Thus, we evaluated the image quality generated under low radioactivity using deeplearning algorithms. In addition, internal dosimetry for various organs was measured based on Monte Carlo simulation. The dataset comprised 500 slices of thyroid nuclear medicine images acquired at 370 MBq as high radioactivity and 18.5 MBq as low radioactivity. The residual neural network (ResNet) with 16 residual blocks and a U-Net model were used with various hyperparameters for learning. In addition, we measured the internal dosimetry in peripheral organs based on a Monte Carlo simulation. The peak signal-to-noise ratio (PSNR) results were 18.24 for the ResNet model and 18.85 for the U-Net model on average. The structural similarity index measurement (SSIM) results were also 0.995 and 0.994 for the ResNet and U-Net models, respectively. The average absorbed dose for radionuclides with 37 MBq was 3.98 times higher than that of the radionuclide with 18.5 MBq. In conclusion, it is possible to make a diagnosis based on the administration of low radioactivity when scanning thyroid nuclear medicine images.

关键词： deep learning architecture Monte Carlo simulation Geant4 application for tomographic emission Thyroid phantom image Nuclear medicine

来源：评论

学校读者我要写书评

暂无评论

Early Detection of Red Palm Weevil in Agricultural Environment Using deep learning

引用

OPTICAL MEMORY AND NEURAL NETWORKS 2025年第1期34卷 63-76页

作者： Sayed, Gehad Ismail Ibrahim, Samar Hassanien, Aboul Ella Canadian Int Coll CIC Sch Comp Sci Cairo 4731261 Egypt Cairo Univ Fac Comp & Artificial Intelligence Giza 12613 Egypt

The red palm weevil (RPW) represents a significant danger to palm trees farms all over the world, which will result in considerable financial losses. The absence of apparent signs until the death of the palm tree makes it difficult to identify RPW infections at an early stage. The prompt detection of RPW diseases is further complicated by large-scale farms. In order to accomplish early detection of RPW using image analysis, this paper proposed a RPW classification model based on the proposed modified ResNet-34 deep learning architecture. A dataset of 483 images is used to assess the model's performance. For the assessment, two different dataset settings are used. In the initial dataset setup, images are divided into three groups: adults, eggs, and Pupae. Four additional categories are added to the classification in the second dataset setup: female adults, male adults, eggs, and pupae. Experimental findings show the usefulness of the proposed model, with a remarkable total accuracy of 98% for both dataset setups. These results highlight the value of using the modified ResNet-34 architecture for the early detection of RPW. Moreover, the findings demonstrated that the proposed model offers great potential for decreasing the negative effects of RPW on palm tree farms and preventing financial losses in the agriculture sector.

关键词： agriculture sustainable development red palm weevil machine learning deep learning architecture

来源：评论

学校读者我要写书评

暂无评论

deep learning modelling techniques: current progress, applications, advantages, and challenges

引用

ARTIFICIAL INTELLIGENCE REVIEW 2023年第11期56卷 13521-13617页

作者： Ahmed, Shams Forruque Alam, Md. Sakib Bin Hassan, Maruf Rozbu, Mahtabin Rodela Ishtiak, Taoseef Rafa, Nazifa Mofijur, M. Ali, A. B. M. Shawkat Gandomi, Amir H. Asian Univ Women Sci & Math Program Chattogram 4000 Bangladesh Asian Inst Technol Data Sci & Artificial Intelligence Chang Wat Pathum Thani 12120 Thailand Carnegie Mellon Univ Dept Computat Biol Pittsburgh PA 15213 USA Carleton Univ Sch Comp Sci Ottawa ON K1S 5B6 Canada Univ Cambridge Dept Geog Downing Pl Cambridge CB2 3EN England Univ Technol Sydney Ctr Technol Water & Wastewater Sch Civil & Environm Engn Ultimo NSW 2007 Australia Prince Mohammad Bin Fahd Univ Mech Engn Dept Al Khobar 31952 Saudi Arabia Cent Queensland Univ Sch Engn & Technol 300 Melbourne Vic Australia Univ Fiji Sch Sci & Technol Lautoka Fiji Univ Technol Sydney Fac Engn & Informat Technol Sydney NSW 2007 Australia Obuda Univ Univ Res & Innovat Ctr EKIK H-1034 Budapest Hungary

deep learning (DL) is revolutionizing evidence-based decision-making techniques that can be applied across various sectors. Specifically, it possesses the ability to utilize two or more levels of non-linear feature transformation of the given data via representation learning in order to overcome limitations posed by large datasets. As a multidisciplinary field that is still in its nascent phase, articles that survey DL architectures encompassing the full scope of the field are rather limited. Thus, this paper comprehensively reviews the state-of-art DL modelling techniques and provides insights into their advantages and challenges. It was found that many of the models exhibit a highly domain-specific efficiency and could be trained by two or more methods. However, training DL models can be very time-consuming, expensive, and requires huge samples for better accuracy. Since DL is also susceptible to deception and misclassification and tends to get stuck on local minima, improved optimization of parameters is required to create more robust models. Regardless, DL has already been leading to groundbreaking results in the healthcare, education, security, commercial, industrial, as well as government sectors. Some models, like the convolutional neural network (CNN), generative adversarial networks (GAN), recurrent neural network (RNN), recursive neural networks, and autoencoders, are frequently used, while the potential of other models remains widely unexplored. Pertinently, hybrid conventional DL architectures have the capacity to overcome the challenges experienced by conventional models. Considering that capsule architectures may dominate future DL models, this work aimed to compile information for stakeholders involved in the development and use of DL models in the contemporary world.

关键词： deep learning deep learning architecture Neural network Boltzmann machine deep belief network Autoencoders

来源：评论

学校读者我要写书评

暂无评论

Mammogram learning System for Breast Cancer Diagnosis Using deep learning SVM

引用

Computer Systems Science & Engineering 2022年第2期40卷 491-503页

作者： G.Jayandhi J.S.Leena Jasmine S.Mary Joans Research Scholar Anna UniversityChennaiTamilnadu600025India Velammal Engineering College ChennaiTamilnadu600066India

The most common form of cancer for women is breast *** advances in medical imaging technologies increase the use of digital mammograms to diagnose breast ***,an automated computerized system with high accuracy is *** this study,an efficient deep learning architecture(DLA)with a Support Vector Machine(SVM)is designed for breast cancer *** combines the ideas from DLA with *** state-of-the-art Visual Geometric Group(VGG)architecture with 16 layers is employed in this study as it uses the small size of 3×3 convolution filters that reduces system *** softmax layer in VGG assumes that the training samples belong to exactly only one class,which is not valid in a real situation,such as in medical image *** overcome this situation,SVM is employed instead of the softmax layer in *** augmentation is also employed as DLA usually requires a large number of *** model with different SVM kernels is built to classify the *** show that the VGG-SVM model has good potential for the classification of Mammographic Image Analysis Society(MIAS)database images with an accuracy of 98.67%,sensitivity of 99.32%,and specificity of 98.34%.

关键词： deep learning architecture support vector machine breast cancer visual geometric group data augmentation

来源：评论

学校读者我要写书评

暂无评论

Comparative analysis of deep learning classifiers for diabetic retinopathy identification and detection

引用

IMAGING SCIENCE JOURNAL 2022年第6期70卷 358-370页

作者： Rayavel, P. Murukesh, C. Sri Sairam Inst Technol Dept Comp Sci & Engn Cybersecur Chennai Tamil Nadu India Velammal Engn Coll Dept Elect & Commun Engn Chennai Tamil Nadu India

Diabetic retinopathy (DR) is a micro vascular problem caused by diabetes that can lead to loss of sight. The early detection of diabetic retinopathy is important to avoid the severity of sightlessness. In this manuscript, a comparative analysis of several deep learning methods for DR identification is proposed. The input fundus images are taken from a standard dataset pre-processed by the Mathematical Morphology process. Moreover, the images are segregated using a Multilevel segmentation of the Region of interest (ROI) based on the split and merge algorithm. After that, an original deep learning architecture is utilized to categorize the segregated fundus images. deep learning methods, such as Convolution neural network (CNN), Recurrent Neural Network (RNN), Support Vector Machine (SVM), Fuzzy K-means cluster (FKM) and Discriminant Analysis (DA) are proposed to classify the DR. The proposed DR identification and detection with CNN provides 65.54% SP, 100% SE, 78.54% SV and 96.95% ACC. Finally, CNN shows better performance than other classifiers.

关键词： Glaucoma categorization segregated pictures mathematical morphology technique split and merge algorithm deep learning architecture

来源：评论

学校读者我要写书评

暂无评论

Vehicle Number Plate Detection using deep learning 2

Vehicle Number Plate Detection using Deep Learning

引用

2nd IEEE International Conference on Integrated Circuits and Communication Systems, ICICACS 2024

作者： Krishna, K.V.S.S. Rama Rajyalakshmi, P. Asritha, K. Lakshmi Divyasri, M. Chaitragna, K. Shanmukha Vignan's Nirula Institute of Technology and Science for Women Department of It Guntur India

ISBN: (纸本)9798350317558

Due to inadequate global population growth, there has been a dramatic increase in the volume of traffic on today's state-of-the-art highways. Vehicle identification and traffic congestion assessment are both critical. The secret is to use readers for license plates to gather data about automobiles. The rapid and precise detection of vehicle crossings required the development of a deep learning framework or networks. The proposed approach relies on the Densenet121 technique for license plate segmentation, with the inclusion of an adaptable attention network to better extract the digits and letters. According to the findings, it is best to use a vehicle predictor powered by deep learning to spot plates and numbers at the same time. To do this, we used Python in conjunction with many deep learning instruments. Experimental assessments have proven that deep learning is effective due to its state-of-the-art detection precision and computational economy. The research also dives into the obstacles encountered and possible remedies for automatic plate recognition in pictures. License plate readers have several practical applications including toll collecting, accident investigation, and identity of suspicious vehicles. © 2024 IEEE.

关键词： conventional neural networks deep learning architecture Densely Connected Convolutional Networks DenseNet121 OCR RCNN TensorFlow

来源：评论

学校读者我要写书评

暂无评论

Music style migration based on generative Adversarial Networks

引用

ALEXANDRIA ENGINEERING JOURNAL 2025年 118卷 292-305页

作者： Ji, Zhen Shen, Dan Hangzhou Vocat & Tech Coll Art Teaching & Res Dept Hangzhou 310018 Peoples R China Yingkou Inst Technol Sch Elect Engn Yingkou 115000 Peoples R China

Music style migration, a key technique in the field of music generation, aims to convert audio from one style to another. However, existing methods often struggle to balance content preservation, style consistency, and sound quality in many-to-many style migration tasks, and are especially deficient in high-frequency detail preservation and complex style mapping. To address these issues, we propose the CAUNet model for efficient many-to-many music style migration by combining the autoencoder, CycleGAN and U-Net augmented generator modules. The autoencoder module extracts the core content features of the audio, the CycleGAN module ensures the authenticity of the style transformation, and the U-Net augmented generator module improves the detail reproduction through multi-scale feature fusion, so that the generated audio meets the target requirements in terms of both style and sound quality. The experimental results show that CAUNet significantly outperforms the mainstream comparison models in terms of content consistency, style matching, and sound quality, and maintains high generation quality in the complex style combination task. The model provides a reliable framework for realizing many-to-many music style migration and has wide application potential.

关键词： Music style transfer Many-to-many generation model deep learning architecture Autoencoder Generative adversarial network Audio detail preservation

来源：评论

学校读者我要写书评

暂无评论

Rotational Photonic Spin Hall Effect Sensor for Liquid Chemicals Classification via EfficientNet-V2

引用

ACS PHOTONICS 2024年第1期12卷 485-494页

作者： Zeng, Kang Zeng, Linzhou Yang, Peng Ke, Yougang Huang, Zhiwei Hunan Inst Sci & Technol Sch Informat Sci & Engn Yueyang 414006 Peoples R China Natl Univ Singapore Coll Design & Engn Dept Biomed Engn Singapore 117583 Singapore Natl Univ Singapore Suzhou Res Inst Suzhou 215123 Peoples R China Natl Univ Singapore NUS Grad Sch Integrat Sci Engn Programme ISEP Singapore 119077 Singapore

Photonic spin Hall effect (PSHE) has been widely used for sensing tasks;however, its potential appears to be unexplored for the development of a compact yet effective sensor for the classification of liquid chemicals. In this study, a liquid identification scheme is demonstrated based on the recently proposed rotational PSHE, where the weak measurement techniques are no longer required for sensing. A liquid crystal device is fabricated to experimentally validate the rotational PSHE, which provides unique beam patterns for liquid analytes. The collected beam pattern images are used to train an EfficientNet-V2-a fast and efficient deep learning architecture-for classifying the liquid chemicals. Two groups of liquids are identified with accuracy over 99% in the proposed scheme. Moreover, the performances of several deep learning models are compared, demonstrating the fast training speed and high parameter efficiency of the EfficientNet-V2. The proposed approach provides an efficient, accurate, and convenient method for refractive index sensing.

关键词： photonic spin Hall effect liquid chemicals identification refractive index sensing liquid crystal element deep learning architecture

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：